Open sourceFree · no sign-up

Try to break an LLM, and see which guardrail stops you.

LLM Guardrails. Test LLM safety with Llama Guard and NeMo Guardrails.

View the code ↗
llm-guardrails
LG
Why it exists

01PROBLEM

LLM apps can be tricked into unsafe outputs, and most teams never test for it.

02SOLUTION

Ready-to-run Streamlit apps for Llama Guard alone, or with Llama 2, plus a NeMo Guardrails setup.

03VALUE

See LLM security risks hands-on before they reach your users.

Screenshots

The story

How to use it

  1. Run the Streamlit app locally.
  2. Try prompts and responses against Llama Guard.
  3. Compare results with and without guardrails.

How it's built

Ready-to-run Streamlit apps using Llama Guard, alone or with Llama 2, plus a NeMo Guardrails setup.

Why it matters for businesses

Unsafe outputs damage trust and create risk. Testing guardrails before launch is far cheaper than fixing incidents after.

In real life

Before launching a customer chatbot, a team tests jailbreak attempts and confirms which ones the guardrail blocks.

Under the hood

How it's built.

// llm-guardrails · Agents & RAG
export const llmGuardrails = {
stack: ["Llama Guard", "NeMo Guardrails", "Streamlit"],
audience: ["AI engineers", "Security teams"],
status: "open-source", price: 0
}
Shape what's next

Feature requests.

No requests for LLM Guardrails yet. Be the first!

Chat with Bala